Including estimates of uncertainty in climate data records

A new paper in Earth System Science Data discusses uncertainty in climate data records (CDRs) from Earth Observation. It emerges from the European Space Agency Climate Change Initiative project, and reviews across 11 different essential climate variables the status of including uncertainty information in CDRs. The paper reviews why uncertainty information matters, what it means (including how it is conceptually different to quality), and briefly reviews how uncertainty can be characterised and quantified. The ECVs included are: aerosol optical depth, cloud properties, glacier extent, greenhouse gases, land cover, ocean colour, ozone, soil moisture, sea ice, sea level and sea surface temperature. The varied nature of these variables leads to a wide range of approaches to uncertainty characterisation, which are summarised.. Nonetheless, the paper pulls together a number of recommendations for good practice that the FIDUCEO project would certainly agree with:

1. Make quantitative uncertainty information available within the dataset. (Don’t expect users to find uncertainty information from reading related papers.)

2. Use well-defined metrological concepts, such as "standard uncertainty", to quantify uncertainty.

3. Provide uncertainty information that discriminates which data are more and less certain. Per-datum uncertainties should be given, if possible, in CDRs where uncertainty varies significantly.

4. Assuming per-datum uncertainty information is provided, avoid redundancy of this information with quality flags:

  • do not flag high-uncertainty data as “bad” if a valid estimate of that high uncertainty is provided;

  • instead, use quality flags to indicate the level of confidence in the validity of the provided uncertainty and retrieval assumptions.

5. Define what uncertainty information is given in the CDR in the product documentation.

6. Describe in the product documentation the main effects causing errors, how uncertainty varies within the dataset, how errors may be correlated in time and space, and under what circumstances estimated uncertainty may be invalid (and flagged as such).

7. Use validation to evaluate both retrieved quantities and uncertainty estimates.

8. Propagate uncertainty appropriately (accounting for error correlation) and consistently when creating aggregated products. 

This represents an overview of the outcomes of many discussions across the different project teams in ESA CCI and is significant progress compared to the level of mutual understanding between teams on how to handle uncertainty at the start of ESA CCI.

However, one aspect central to FIDUCEO is missing from the paper entirely: in CCI, the project teams by necessity have to add uncertainty information at level 2 (geophysical retrievals) without access to per-datum quantified uncertainties at level 1 (the radiances etc from which the climate data are inferred). This is a fundamental limitation of the current state of the art, and a key component of FIDUCEO is to show how characterisation of the uncertainty and error correlation structure at radiance level can usefully inform the creation of uncertainty estimates in derived climate data records. The paper is in discussion at ESSD and comments can be made online as part of open peer review.

Add new comment